Hybrid Approach for Noise Removal and Image Enhancement of Brain Tumors in Magnetic Resonance Images

Abstract

In medical image processing, Magnetic Resonance Imaging (MRI) is one of significant diagnostic techniques. It provides high quality of important information about the analysis of human soft tissue when measured with CT imaging modalities; hence it is suitable for diagnosis at best. However, if it gives quality of information, image may distorted by noise because of image acquisition device and transmission. The noises in MR image reduces the quality of image and also damages the segmentation task which can lead to faulty diagnosis. Noises have to reduce at the same time there is no information loss. This paper propose a hybrid approach to enhance the brain tumor MRI images using combined features of Anisotropic Diffusion Filter (ADF) with Modified Decision Based Unsymmetric Trimmed Median Filter (MDBUTMF). ADF scheme provides a superior performance by removing noise while preserving image details and enhancing edges. MDBUTMF helps in image denoising as well as preserving edges satisfactorily when the noise level is high. The performance of this filter is evaluated by carrying out a qualitative comparison of this method with other filters namely, ADF filter, Modified Decision Algorithm, Median filter, MDBUTMF.